outcome measures in total hip replacement and lumbar spine surgery
Department of Orthopaedics Institute of Clinical Sciences
Sahlgrenska Academy University of Gothenburg
Sweden
Ted Eneqvist, MD
2018
© Ted Eneqvist 2018 ted.eneqvist@gmail.com
The copyright of the contents of this thesis belongs to Ted Eneqvist.
The published/accepted articles are reproduced with permission from the respective journals.
Typeset by Team Media Sweden AB
Cover illustration by Pontus Andersson, Pontus Art Production Printed in Gothenburg, Sweden 2018
BrandFactory AB
ISBN: 978-91-629-0509-5 (PRINT) ISBN: 978-91-629-0510-1 ( (PDF) http://hdl.handle.net/2077/55396
List of papers ...5
Abbreviations ...6
Abstract ...7
Background ...7
Objective ...7
Patients and methods ...7
Results ...7
Conclusion ...7
Introduction ...9
Background ...11
Total hip replacement ...11
Low back surgery ... 12
Hip-spine syndrome ... 15
Orthopedic registers ... 16
Swedish Hip Arthroplasty Register ... 18
Swespine ... 18
Importance of register data quality ... 19
Prospective observational studies ... 19
PROMs ... 19
Other measurements used in these studies ... 22
Aims ... 23
Patients and methods ... 25
Patients ... 25
Ethical considerations ... 26
Methods ... 26
Statistical methods ... 31
Summary of papers ... 33
Paper I ... 33
Paper II ... 33
Paper III ... 36
Paper IV ... 37
Paper V ... 38
Paper II ...42
Paper III ...42
Strengths and limitations ...45
Discussion...47
General discussion ...47
The clinical assessment of PROMs following THRs and LSSs ...47
The hip-spine syndrome -where to start with surgery, the hip or the spine? ...48
Conclusions ...53
Future projects ...55
The “hip-spine syndrome” ...55
The “hip-knee syndrome” ...55
Detection of patients with a high risk of reoperation ...55
Sammanfattning på svenska ...57
Project collaborators ...59
Supervisors ...59
Collaborators ...59
Acknowledgements ... 61
References ...63
I. Low back surgery prior to total hip replacement is associated with worse patient-reported outcomes
Eneqvist T, Nemes S, Brisby H, Garellick G, Fritzell P, Rolfson O.
Bone Joint J. 2017 Jun;99-B(6):759–765.
II. Patients with a previous total hip replacement experience less reduction of back pain following lumbar back surgery
Eneqvist T, Bülow E, Nemes S, Brisby H, Garellick G, Fritzell P, Rolfson O.
J Orthop Res. 2018 Apr 12. [Epub ahead of print]
III. Does the order of total hip replacement and lumbar spine surgery influence patient-reported outcomes?
Eneqvist T, Nemes S, Brisby H, Garellick G, Fritzell P, Rolfson O.
In Manuscript
IV. Can patient-reported outcomes predict reoperations after total hip replacement?
Eneqvist T, Nemes S, Bülow E, Mohaddes M, Rolfson O.
Int Orthop. 2018 Feb;42(2):273–279.
V. How do EQ-5D-3L and EQ-5D-5L valuations compare in a Swedish total hip replacement population?
Eneqvist T, Nemes S, Kärrholm J, Burström K, Rolfson O.
In Manuscript
List of papers
This thesis is based on the following papers:
Abbreviation Definition
C Concordance index
CT Computed tomotography
EQ EuroQol
HRQoL Health-related quality of life
LBP Low back pain
LSS Lumbar spine surgery
LSSS Lumbar spinal stenosis surgery
MCID Minimal clinical important difference
MDC Minimal detectable change
MID Minimal important difference
MRI Magnetic resonance imaging
OA Osteoarthritis
ODI Oswestry Disability Index
OLS Ordinary least squares
PIN Personal identity number
PRO Patient reported outcome
PREM Patient-reported experience measurements PROM Patient-reported outcome measurement
RCT Randomized clinical trial
SD Standard deviation
SHAR Swedish Hip Arthroplasty Register
THR Total hip replacement
TTO Time trade off
VAS Visual analog scale
Abbreviations
Abstract
Patients and methods
For Paper I-III, data including PROMs on patients with THR and LSS performed in 2002–2012 were obtained from SHAR and Swespine and linked to identify those who occurred in both registers. In Paper IV, data from SHAR on patients with THR in 2002–2014 were used to establish the relationship between PROMs and reoperation. For Paper V, patients eligible for THR in western Sweden during 2015 were invited to answer EQ-5D-3L and EQ-5D-5L with a two-week separation before and after surgery. Logistic and linear regression analyses were used to investigate research questions.
Results
Patients with both THR and LSS performed had worse one-year PROMs following the last procedure compared to patients with surgery in only one location. Patients eligible for both THR and LSS within a short period of time had better outcomes following the last procedure if surgery started with LSS. PROMs collected one year following THR predicted the risk of subsequent reoperation. Patients frequently utilized the additional response options of EQ-5D-5L and ceiling effects at the one-year follow-up were reduced compared to EQ- 5D-3L. EQ VAS estimates for different severity levels conformed well between questionnaires.
Conclusion
This thesis contributes to the understanding of patient- reported outcomes for patients who undergo both THR and LSS. Given their ability to predict reoperations following THR, PROMs can be utilized to identify patients at increased risk, which may may be used to improve follow-up routines and care. Since EQ-5D-5L better describes health-related quality of life in THR patients, the introduction of the extended questionnaire as a standard tool in SHAR will enable a more accurate assessment of the procedure.
Background
Beginning in the late 1990s, the Swedish Hip Arthroplasty Register (SHAR) and the Swespine have successfully implemented programs to collect patient-reported outcomes measures (PROMs). The use of PROMs has enabled assessment of patients’ health-related quality of life (HRQoL), physical function and pain following total hip replacement (THR) and lumbar spine surgery (LSS). The nationwide collection of PROMs has made it possible to evaluate changes of care, compare providers, investigate factors influencing outcomes that matter for patients, and it has contributed to improvement in clinical practice.
Objective
The overall objective of this thesis is to investigate different ways to utilize PROMs following total hip replacement and lumbar spine surgery. Specifically, this thesis aims to:
• Investigate PROMs in patients who have undergone LSS prior to THR and in patients who have undergone THR prior to LSS compared to matched patients with isolated THR or LSS.
• Investigate if the order of THR and LSS affects PROMs one year following the last procedure in patients with both procedures performed within a period of two years.
• Investigate if PROMs can predict the risk for reoperation following THR.
• Assess the measurement properties of EQ-5D- 5L compared to EQ-5D-3L in a Swedish THR population and to estimate how different severity levels of the two versions of the questionnaire conforms.
For many years to come the number of primary total hip replacements and lumbar spine surgeries will continue to increase, not only in Sweden but also in a global perspective (1, 2). Several factors are able to explain this increase. Improved implant technology and surgical techniques make surgical procedures available for both older and younger patients. Due to improvements in general medical practice and educational level, the general health is improving resulting in increased life span, an increasing population and increased demand for these surgical procedures.
Both hip osteoarthritis and spinal stenosis are common degenerative diseases in the general population.
Subsequently, the concurrence of these musculoskeletal disorders is commonly encountered in clinical practice.
Hip osteoarthritis and spinal stenosis often present with similar symptoms, which may make it difficult to determine the origin of pain. The “hip-spine syndrome” was first described by Ofierski et al in the late 1970s. ref (72–75) Due to the increasing life span of the population, more and more people are likely to need total hip replacement (THR) and lumbar spine surgery (LSS) in one or both locations over time. In the presence of concurrent symptomatic conditions in the hip and lumbar spine, the location with which to begin has long been the subject of debate. Knowledge about the outcome in patients in whom both procedures are performed is limited.
For patients with hip osteoarthritis where non-surgical treatment is ineffective, total hip replacement is a well- established and cost-effective treatment. The survival of some prosthetics has been reported to be above 95% at 10-years follow-ups. Since complications are uncommon after THR, few patients require reoperation following a standard THR procedure. Most providers of joint replacement surgery have abandoned regular follow-ups after routine THRs. Unfortunately, an
Introduction
important minority operated with THRs will experience complications, and some will need reoperation due to early or late complications. An event of this kind might be devastating for the individual patient, often resulting in impaired function and disability (3–5). In addition, these events are very costly to society and the health care system. The opportunity to detect patients running a higher risk of reoperation could possibly reduce the suffering and costs associated with patients with complications following THR.
Patient-reported outcome measures (PROMs) are essential tools in the assessment of outcomes following THR procedures (6). One of the most commonly used health-related quality of life (HRQoL) instruments is the original three level form of the EQ-5D, the EQ-5D- 3L (7, 8). However, the EQ-5D-3L has been questioned due to its low sensitivity and the lack of descriptive richness, which has been shown following THR in several studies (9, 10). In response to this critisism of the EQ-5D-3L, a five level version of the questionnaire has been developed, the EQ-5D-5L, which expands the range of responses in each of the five dimensions from three to five levels (11). However, the usefulness of the 5L version has not yet been established among Swedish THR patients.
Papers I-III presented and discussed in this thesis aim to use PROMs to explore the outcome following surgery in patients who undergo both THR and LSS and to investigate differences in patient-reported outcomes depending on the order in which surgeries were performed. Paper IV aims to investigate whether PROMs one year postoperatively are able to predict the patients who run a higher risk of reoperation. The last paper (V) compares the new version of the PROM instrument, the EQ-5D-5L, with the EQ-5D-3L in a Swedish total hip replacement population.
Total hip replacement
Hip osteoarthritis (OA) is a common degenerative disease primarily affecting the aging population. In Sweden, the prevalence has been estimated at 10% in those over 85 years but only at 1% in patients under 55 (12). Other reports indicate a significantly higher prevalence of hip OA; 27% in individuals over 45 on plain radiographs (13).
The natural course of hip OA is progressive with the loss of cartilage in both the caput femoris and acetabulum.
As the chronic and irreversible degenerative progress continues, subchondral cysts and osteophytes may develop. Typical symptoms of hip OA are pain on activity and stiffness of the affected hip. However, pain generated from the hip may present in different ways, such as groin pain, radiating pain down the leg towards the knee, buttock pain, or pain radiating towards the back. The appearance of pain can also differ. Some patients experience only pain on movement, while others have pain at rest, or even sleep-depriving pain. The majority of patients also often suffer from reduced mobility, caused by pain and stiffness.
Diagnostics
Since hip OA presents with different symptoms, diagnosing needs to be individualized to some extent.
Symptoms of hip OA may for instance be present with or without minimal radiographic findings. On the other hand, radiographic findings may be present without causing symptoms. Different sources of information must therefore be considered for the assessment. They include the anamnestic information on symptoms, plain radiographs and/or other imaging technology. There are several other disorders, such as degenerative lumbar diseases or knee OA that can mimic hip OA symptoms, and this should be taken into account before making a diagnosis of hip OA.
Treatment
The first line of treatment is non-surgical including physiotherapy, analgesics and information. If non- surgical treatment fails, the alternative is surgery. In 2016, 17,000 THRs were performed in Sweden, and the vast majority were due to OA. This surgical treatment of hip OA consists of irreversible replacement of the femoral head and the acetabulum with an artificial ball-and-socket joint. Sir John Charnley pioneered this procedure in the 1960s. The excellent post-operative results and the cost effectiveness of THR, this procedure has been regarded
Background
as “the operation of the century” (14, 15). Several studies and arthroplasty registers have reported the survival of some prostheses at above 95% at 10 years (16, 17). Even if it is called “the operation of the century”, it is not a risk- free procedure. A systematic review concluded that the average 90-day mortality after THA was 0.7% (18). THR for hip OA is an elective procedure to relieve pain and improve mobility and HRQoL. As a result, expectations of the outcome are reasonably high. Hip OA is not a life-threatening diagnosis, THR should therefore only be offered to those who will most likely benefit from the procedure.
Reoperation of total hip replacement surgery
Although the majority of THR patients will live the remainder of their lives without requiring further surgery, some patients will need a reoperation due to early or late complications. These complications could be loosening, infection, fracture or dislocation of the prosthesis (16). Every year in Sweden, about 2200 reoperations are performed on patients with THR (16). The SHAR defines a reoperation as any further surgery performed on the hip. A revision is a reoperation in which any implant components are exchanged, removed or added.
Examples of reoperations that are not revisions are open reduction, soft-tissue repair, debridement without exchanging implant parts, the removal of heterotopic bone formation and the relief of hemorrhage. Generally, reoperations result in impaired function and disability and constitute a risk factor for repeat surgical interventions (3–5). The result following revision surgery are less likely to be as successful as the first operation (19). In addition, these events are very costly, in terms not only of medical costs but also of the costs of loss of productivity related to patients’ loss of working capacity. It is therefore of the utmost importance to minimize the risk of these events.
The need for routine follow-ups and prediction of reoperation
As the incidence of primary THRs is rising and complications are uncommon following THRs, few patients need a reoperation following a standard THR.
The increasing numbers of THRs produces an increasing number of follow-ups, at an increasing cost. Those complications that do occur following THRs are often of an acute nature and cannot be foreseen on a return visit. The need for follow-ups after standard THRs for all patients has therefore been questioned (20, 21). As a result, many
health-care providers have abandoned routine follow- ups after standard THRs. However, some complications, such as loosening of the prosthesis, may be symptomatic and could possibly be detected at follow-up visits. Some complications require less extensive interventions if they are detected at an early stage. Follow-up visits may also constitute an important function in identifying issues such as degenerative spinal conditions, contralateral hip disease, poor strength, limited hip range of motion and impaired walking ability eligible for physiotherapy or further surgical interventions. Although routine follow-ups for all patients are not justifiable, it is desirable to develop measures to identify patients running a higher risk of complications, so they can be called to follow-ups. Considering the extensive research on risk factors for reoperation following THR, it is feasible to establish methods to identify high-risk patients.
These instruments could potentially include patient- related factors (22–25), implant- and surgical procedure- related factors (24, 25) and PROMs (26–28). Automated and individualized assessment of variables associated with the risk of reoperations could be used to detect patients that would potentially benefit from monitoring. However, no models that have a specificity high enough to be used in a clinical setting are as yet available.
Low back surgery
Lumbar spinal stenosis, spondylosis/spondylolisthesis and segment-related pain are degenerative low back diseases primarily affecting the aging population. LSS is the most common of these diagnoses and its prevalence has been estimated at up to 60% among patients over the age of 60 on magnetic resonance imaging (MRI) (29). The prevalence of these diagnoses is increasing in Sweden, just as it is in the rest of the developed countries.
The reason for this increasing trend is probably that these diagnoses are being more commonly diagnosed among the population. This is probably due to multiple reasons. In part, it could reasonably be explained by a change in the demographic situation, with an older age structure among the population, and that symptoms of these diagnoses are not accepted as before, since the population imposes increasingly higher demands on physical activity at an older age. Another reason for the increasing trend for these diagnoses could perhaps be the increasing availability of MRI. In some cases, lumbar spinal stenosis will require surgical intervention as a final treatment (30). In Sweden, the annual rate of surgery due to these diagnoses is close to 55–60 per 100,000 inhabitants and in the past decade there has been a yearly increase of 5–10% according to the Swedish Spine Register, Swespine (2, 31). In the US, 14 per 10,000
inhabitants over the age of 65 undergo lumbar spinal stenosis surgery (LSSS) each year, which represents a fourfold increase since 1985. LSSS has become the most common indication for spinal surgery in Sweden and in other European countries (31–34).
Symptoms and diagnosis
The symptoms of lumbar spinal stenosis are caused by a narrowing of the spinal canal, compressing the neural structures as the nerve roots leaving the spinal cord. This narrowing is caused anteriorly by a bulging disc, dorsally by a thickening of ligamentum flavum and laterally by osteophytes derived from osteoarthritic changes in the joint facets (35, 36). Another contributory factor to stenosis can be degenerative spondylolisthesis. This is caused by an ongoing degenerative process of the lumbar spine segments that can lead to a forward slip of the adjacent upper vertebra (37). The symptoms of lumbar spinal stenosis and other degenerative disorders affecting the lumbar spine are characterized by back pain, numbness, or radiating pain to the buttocks and lower extremities, as well as muscle weakness. At a later stage, walking disability may also appear (38, 39). The symptoms are often referred to as pseudoclaudicatio or neurogenic claudication, and need to be distinguished from genuine claudicatio or claudicatio intermittens caused by arterial insufficiency. Lumbar spinal stenosis symptoms may mimic pain caused by hip and knee joint disorders or polyneuropathy. These differential diagnoses should be taken into consideration when determining the cause of symptoms. As the conditions causing lower limb pain often coincide and contribute individually to the range of symptoms a patient may experience, it is a diagnostic challenge to differentiate the origin of the different pain components. While anamnestic information is of the greatest importance in diagnosing these disorders, a diagnosis of lumbar spinal stenosis requires image-technology confirmation. MRI is now the method of choice and should be used, taking absolute and relative contraindications (for example a pacemaker, ear implant, or claustrophobia) into consideration. The grade of lumbar spinal stenosis can be determined by using the Schizas 7-grade classification which is based on the morphology of the dural sac as observed on MRI based on the rootlet/cerebrospinal fluid ratio (40). The grade of lumbar spinal stenosis can also be measured with a cross section of neural structures of the spinal canal were75mm2 or less has been shown to correlate with clinical symptoms, and is considered a confirmation of lumbar spinal stenosis (41). If MRI is contraindicated a combined examination with CT (computed tomography) and myelography can be performed to make a correct diagnosis of lumbar spinal stenosis.
Surgical and non-surgical treatment
The treatment of degenerative diseases of the lumbar spine such as spinal stenosis may consist of physical activity and patient education, epidural joint injections, or surgical intervention. There is no strong evidence of any long-term effects of steroid epidural joint injections (42). Physiotherapy has been shown to postpone surgery by four to six months (43, 44), but there are no studies presenting physical treatment as a long-term solution. There is evidence that surgery is a better long-term alternative compared with non- surgical treatment (45–48). The surgical treatments of these degenerative diagnoses of the lumbar spine is decompression or decompression and fusion of the affected segment. Decompression aims to ease pressure on the neural elements in the stenotic spinal canal. Since the description of pedicle screws by Roy-Camille et al.
in 1970 (49), the use of the pedicle instrumentation of affected segments to achieve fusion has gradually increased and it is now used routinely in spinal surgery.
Fusion vs non-fusion
The significance of instrumentation with pedicle screws in relation to the healing rate of spinal fusion has been a subject of debate. Two studies reported no differences
in healing rates of instrumented and non-instrumented fusions (50, 51). However, other research reported significant increases in fusion healing for instrumented fusions in clinical and animal studies (52–54). A Cochrane review from 2006 claimed strong evidence of higher fusion rates for instrumented fusion compared with non-instrumented fusion (55). It has been debated whether decompression causes instability of the spine (56, 57). To prevent instability, it has therefore been customary to perform fusion as a complement to decompression, particularly in the presence of degenerative spondylolisthesis. However, according to a recent randomized clinical trial by Försth et al., fusion as a complement to decompression does not lead to better post-operative results or cost-effectiveness compared with decompression alone. These findings were true even in the presence of degenerative spondylolisthesis preoperatively (58, 59). In a study of biomechanics, Försth et al. found that the potential instability caused by decompression was minimal and removal laminectomy did not result in increased instability compared with bilateral laminectomy (60). They suggested that the main principle of LSSS should consist of decompression alone, even if degenerative spondylolisthesis is present pre-operatively.
Normal
Lumbar disc
Degenerative changes Compressed nerves Spinal canal
Spinal stenosis
Figure 1. Pictures presenting normal spinal canal and spinal stenosis.
L4
B. Decompressive laminectomy A. Incision
C. Foraminotomy and facetomy D. Fixation
Figure 2. Pictures presenting decompressive and fusion surgery.
Complications following LSSS
Surgical procedures for lumbar spinal stenosis are considered safe, with a reported mortality within 90 days of 0.3–0.4% (61–63). However, there are other risks following LSSS. In a systematic review, the risk of surgical complications (nerve root injury, cauda equinae damage, bleeding in the spinal canal, dura lesion and wound infection) were reported in approximately 5–10%
and general medical complications (anaesthesiological, cardiovascular, pulmonary, cerebral, kidney/urinary, and liver/GI) were reported in about 3% of cases (64). A dura tear resulting in cerebral spinal fluid leakage has been reported at even higher levels of 2.1–16% (65–
69). Lumbar spinal stenosis surgery is not a lifesaving procedure, the goal is to maintain or increase activity levels, relieve pain and restore health-related quality of life (HRQoL). The decision to proceed with surgery should therefore only be made if the patient has a substantial impairment and conservative treatment fails to control symptoms.
Hip-spine syndrome
Up to 27% of patients over 45 years have signs of hip OA on plain radiographs (13) and 60% of patients over 60 have signs of lumbar spinal stenosis on magnetic resonance imaging (MRI) (29). Due to the high prevalence of these degenerative diseases of the hip and lumbar spine, it is not surprising that the prevalence of coexisting degenerative lumbar spine disorders has been estimated at up to 18% among a THR-population (70). Similarly, among patients eligible for lumbar back surgery, the prevalence of osteoarthritis of the hip or knee has been reported at 10% (71). The combination of degenerative disorders of the hip and lumbar spine is known as the “hip-spine syndrome” and was first described by Ofierski et al. in 1976 (72–75). These degenerative diseases may cause similar symptoms, which make determining the origin of pain in clinical practice and assessing of how these conditions contribute to symptoms problematic, as well as differentiating them from other diseases with similar symptoms, such as arterial insufficiency and polyneuropathy.
Surgical outcome
In general, lumbar spine surgery and hip replacement are well-investigated interventions and the evidence of their effectiveness in reducing pain and restoring mobility is comprehensive. However, there is some variation in outcomes and a great deal of research has focused on exploring factors associated with better and worse results. Increased comorbidities affecting mobility
and combinations of degenerative musculoskeletal disorders have been shown to be associated with poorer outcomes following surgery. Comorbidities are defined as patient conditions or diseases not associated with the development or cause of the immediate disease of interest. Comorbidities can be diagnosed at different points in time, which may lead to different associations with the risk of adverse outcomes (76, 77). Since musculoskeletal comorbidity is associated with worse patient-reported outcomes following THR (78), patients with hip-spine syndrome are expected to have poorer outcomes after THR compared with patients with an isolated degenerative hip disorder.
There are reports of low back pain prior to THR being associated with poorer outcome and function following surgery (79, 80). Patients with a known degenerative disease in their lumbar spine have worse outcomes in terms of function, physical-status, activity-levels and satisfaction after THR (81). Furthermore, THR patients who had been diagnosed with lumbar spine disorders experienced less improvement in function and pain compared with patients without a history of lumbar spine disorder (5). Similarly, several studies report poorer outcomes following LSS in the presence of preoperative conditions affecting walking capacity, such as hip osteoarthritis (71, 82–85). Reversely, LSS with the absence of comorbidity and disorders affecting walking capacity prior to LSS are associated with better functional outcomes and pain relief (85, 86). However, the knowledge of the outcome following THR and LSS in patients with both procedures performed is limited.
Where to perform surgery first: the hip or the spine?
For years, there has been an ongoing discussion about whether to perform surgery on the hip or spine for patients with hip-spine syndrome. Some argue that THR should be performed first because of its reliable effectiveness (74). One argument in favor of starting with THR is its association with reduced back pain and the fact that residual symptoms may be effectively treated with subsequent lumbar spine surgery (70, 73–74, 87–90). Others favor LSS initially because of the risk of contraction of the spinal nerve roots during the THR procedure, which may cause nerve damage (91, 92).
Previous research is therefore inconclusive and somewhat conflicting.
Pain caused by degenerative disease in the hip and lumbar spine
Pain is a complex and subjective experience that could be suspected to affect everyone at some time during a life span. When discussing pain from hip OA and spinal stenosis, both acute and chronic pain needs to be considered, as well as pain arising from nociceptive- and neuropathic pathways.
Acute and chronic pain
Physiologically acute pain is meant to serve as an internal alarm system helping us to react to external dangers in our environment or to internal dangers caused by potentially harmful changes in our bodies.
Chronic pain, on the other hand, is always maladaptive and the continuous somatosensory and emotional burden can seriously affect the sufferers quality of life (93). Although chronic pain is a characteristic of many different diseases, hip OA and low back pain as a part of spinal stenosis are two of the most predominant, putting them among the leading causes of disability worldwide (93, 94).
Nociceptive and neuropathic pain
Nociceptive pain is defined as pain arising from actual or threatening damage to non-neural tissue and it is due to the activation of nociceptors (95), or as pain attributable to the activation of the peripheral receptive terminals of primary afferent neurons in response to noxious chemical, mechanical, or thermal stimuli (96). For clinical purposes, the term nociceptive pain can be used when pain is proportional to nociceptive input, and the term is designed to contrast with neuropathic pain. The latter is defined as pain caused by a primary lesion or disease of the somatosensory nervous system (95). Pain in hip OA is mostly caused by nociceptive pain, but up to 23% of patients also have neuropathic pain (97). Within the low
back pain population lumbar radiculopathy is a common type of lumbar neuropathic pain, while myofascial tissue (i.e. thoracolumbar fascia) (98) and some lumbar ligaments (99) contain nociceptors capable of generating nociceptive pain. For patients with lumbar spinal stenosis, neuropathic pain have in general been reported to be present in more than one third, and for those patients with radicular pain neuropathic pain has been reported to be present in more than two thirds (100). Because of the high concentration of nociceptors in somatic tissues, chronic somatic pain is typically well localized and often results from degenerative processes (such as arthritis).
However, around 15–25% of patients with chronic pain are thought to have neuropathic pain (101).
Pain referral pattern
Pain arising from both osteoarthritis of the hip and lumbar spinal stenosis has complex pain referral patterns. Pain arising from both locations has been shown to be present in several locations other than just the hip or spine. There are several studies describing the pain referral patterns from both the lumbar spine (95, 102) and hip OA (103, 104). The distribution patterns for the pain are similar in, for example, the buttocks, anterior and posterior thigh, groin, low back and knee.
Hip OA has previously been thought not to radiate below the knee and that groin pain is associated with hip OA. However, these symptoms have been proved to originate from both locations (104). Due to the difficulty involved in determining the origin of pain, diagnostic tests such as intraarticular injection of the hip joint or spinal nerve root block with local anesthetics have been recommended (75, 91, 102, 105, 106).
Orthopedic registers
National prospective observational registers
National quality registers have three main objectives: to monitor outcomes, stimulate to improvement activities and to facilitate research. Since the SHAR registers type of the prosthesis used in detail, implant surveillance is a fourth important objective. The national quality registers enables studies to determine the demography of the population with certain conditions or undergoing a specific procedure. The opportunity to obtain large sample sizes provides high statistical power. Together with the continuous validation of these registers, this provides for high reliability in the analyses of survival and outcome of implants and techniques. The large sample sizes and repeated validation processes in these registers enable certain confounders to be taken into consideration when determining whether differences
in outcomes are related to the implant or the technique in question. There are naturally data entry errors in all registers, but these have been proven to be minimal (16, 107). Because of prospective collection and continous validation procedures recall bias have been significantly reduced. When the registers are used for prospective observational register studies, the large sample sizes reduce selection bias and significantly reduce confounding bias and heterogeneity.
Stepwise introduction of new implants
Registers monitoring implant survival such as the SHAR also play an important role in the introduction of new prostheses and surgical techniques. It is preferable for the introduction of new technologies or techniques to be performed step-wise, starting with a small cohort which is closely observed to determine eventual early complications. This step is followed by larger multicenter studies and that new technologies or techniques are finally investigated in large scale prospective observational register studies (108). A step-wise introduction of this kind has the potential to eliminate failure at an early stage and prevents the premature introduction of new prostheses, technologies and techniques on a nationwide basis. It is, however, important to consider the last step in the introduction of observational register studies, which are by nature not designed to explain causation. Causation is determined through cohort and randomized controlled trials where causation can be determined.
PROMs and registers
Since their development, observational orthopedic registers have been used to monitor the survival of implants using revision as the endpoint. By doing this, the success of a
prosthesis has been based upon the survival of the implant, the functional status performance assessed by the surgeon and plain radiographs. When measuring success in this way, interest focus not on the patient´s opinions of the outcome but on the surgeon’s opinion of the outcome following surgery. The opinion of the patient is however most important, since the main indication for the procedure is pain and disability. However, in the last few decades this evaluation approach has shifted to measuring patient- reported outcomes. Ultimately, patients are uninterested in their x-rays, or whether their walking ability as assessed by the surgeon is judged as good, if they are still in pain and dissatisfied with the outcome of surgery. Patient-reported outcome measures (PROMs) provide another dimension to success following orthopedic surgery. PROMs measure not just pain and function but also the patients HRQoL, which can be used to calculate the cost effectiveness of procedures and techniques. PROMs also provide the opportunity to measure outcomes at both group and individual levels. Together with demographic and surgery-specific variables, PROMs enable further register developments aimed at optimizing the care of patients in need of orthopedic surgery.
Linkage of register databases
Personal identity numbers (PIN) are used as a common identifier in most registers in Sweden, in both orthopedic registers, such as SHAR or Swespine, and in other national health data sources, such as Statistics Sweden or the Prescribed Drug Register. This enables the linking of registers, which in turn offer an important opportunity not only to study and adjust for a number of confounders but also to add other outcome measures such as adverse events and sick leave.
The personal identity number
In 1947, the Swedish government introduced the personal identity number (PIN). The personal identity number is a unique 10-digit number. The first six digits contain the birth date (year, month and day). A hyphen separates them from the serial number made up of the three digits, where the third digit identifies the sex, odd numbers for males and an even number for females.
The tenth and last digit is a control number. All citizens in Sweden have their own unique PIN. Since this has been used in the majority of health registers, research by register data has been very successful in Sweden.
Using a personal identity number makes it possible to link data from different registers at individual level. By doing this, registers complement one another with data by reducing confounding and enabling more complex research questions.
Swedish Hip Arthroplasty Register
The Swedish Hip Arthroplasty Register (SHAR) began in 1979. Its purpose was to gather prospective observational data on all hip replacement surgery in Sweden at both publicly and privately funded hospitals. The data that are collected are used to compare results across providers and to monitor longitudinally outcomes of the procedures with the emphasis on implant types, surgical techniques, and complications. As the SHAR collects PROMs data continuously on the Swedish THR population both before and at one, six, and ten years after surgery, the register is able to attain a higher statistical power than a randomized controlled trial or data collected at a single hospital. Due to the large size and complete coverage, the register is also able to quickly identify complications associated with both implants and surgical techniques.
Since 1992, all orthopedic clinics, both public and private, have reported their primary surgeries and subsequent reoperations to the SHAR. In the SHAR coverage is 100%, and the completeness of THR registrations has been reported as 98.1 % (16, 107).
PROMs in SHAR
In 2002, the SHAR launched a nationwide PROMs program for elective THR patients. The program reached full participation among Swedish orthopedic hospitals in 2008. The purpose of the program was to complement the traditional outcome variables, such as implant survival, with patient-reported outcomes relating to pain, function and health-related quality of life (6, 77). The PROMs program invites all patients scheduled for elective THR to participate. Patients are asked to complete a short questionnaire at their pre-
operative visit. A follow-up survey to be answered manually is mailed to patients at one, six, and ten years post-operatively. Every month, the SHAR centrally distributes lists of patients that are to receive follow- up questionnaires to the orthopedic departments where specially trained secretaries at the departments are responsible for sending out questionnaires and reminders and entering data in the on-line PROM database. Response frequencies have been reported as 86% pre-operatively, and 90% at the one-year follow-up (6). The PROMs program comprises the EQ-5D health status questionnaire (109), a hip pain visual analog scale (VAS) (110) and at follow-up a VAS addressing outcome satisfaction. Additionally, the patient’s musculoskeletal comorbidity is determined using the Charnley classification (111). The PROMs program in the SHAR is the voice of all patients operated with a THR. This information provides an excellent opportunity for developing improved healthcare and providing better outcomes for patients operated with a THR.
Swespine
The Swedish Spine Register (Swespine) was started in 1993. Its aim was to prospectively gather observational data on all surgical procedures on the spine performed in Sweden from both publicly and privately funded hospitals.
As in the SHAR, the collected data are used to compare results across providers and to monitor longitudinally the outcomes of the procedure, focusing on surgical techniques and complications. The surgeon’s contribution to the register is to make the diagnosis and classification, together with details of the surgical technique, implants and perioperative complications. The completeness of registrations to the Swespine has been reported as 85%
(2). The unique personal identity numbers (PIN) given to all inhabitants in Sweden are used as identifiers.
PROMs in the Swespine
The PROMs included in the Swespine comprises questionnaires containing the EQ-5D-3L (112), the Oswestry Disability Index (ODI) (113), back and leg pain according to a visual analog scale (VAS) (110), and until recently the Short Form 36 (abbreviated health survey, SF-36) (114). Questionnaires are filled out at the pre-operative visit, or sent to patients prior to surgery.
The same questionnaire, together with a questionnaire regarding satisfaction with the treatment, is filled out at follow-ups one, two, five, and ten years following surgery. The questionnaires are unrelated to any hospital visit, and are completed without the assistance of the surgeon or any other person involved in the treatment.
The Swespine distributes lists of patients due to receive
follow-up questionnaires to the orthopedic departments and specially trained secretaries at the departments are responsible for sending out questionnaires and reminders and entering data in the online Swespine database.
Importance of register data quality
The quality of a register analysis or prospective register studies using register data depends largely on the quality of the data in the current registers database. There are five dimensions in particular that need to be take into consideration; validity and reliability, coverage and completeness and response rates. Both the SHAR and Swespine are continuously working on data validation.
The SHAR does this by examining medical records from all reoperations. Once a year, all orthopedic departments are also requested to compare the register´s numbers with each local hospitals’ patient administration system. The register online entry application also has a built-in warning system for incorrect entries such as the wrong PIN, paired side and implants. When it comes to the reliability of a register it is important that the variables that are included have high consistency or precision of the measuring instrument. The coverage of a register is simply calculated as the proportion of participating units compared with all units, where a high number is good. There are pitfalls to take into consideration, if the coverage is calculated as a possible coverage and not an actual coverage. The coverage is misleading and not a “true coverage”, that should be on an individual procedure level. Completeness depends on that the respective participating unit’s report at individual level. Both the SHAR and Swespine collaborate with the Swedish National Board of Health and Welfare which operates the National Patient Register based on PINs. Departments are required by law to report all medical interventions to the Patient Register. Each year a linkage between the Swedish Hip Arthroplasty Register and the Swespine respectively with the National Patient Register is performed. In this way registers completeness can be calculated at both departmental and individual level.
The figures are published each year in respective registers Annual Reports and this has led to some out-liers rapidly improving their registration. Poor completeness may lead to flawed analyses and feedback will then be misleading, so, if some departments have low completeness they have usually improved their registration.
Prospective observational studies
Prospective observational studies such as register studies obtain data from groups who have or have not been exposed to the subject of interest. In these studies
there are no interventions exposure by the researcher or anyone else. Prospective register studies are preferable when investigating the effects of predictive risk factors on an outcome. To study the effects of an intervention, randomized clinical trials (RCT) are the gold standard. In an RCT, the participants are assigned to either intervention or control/placebo, preferably using a blinded random selection. There are advantages and disadvantages to both types of study. In an RCT the advantages are the unbiased distribution of confounders, the opportunity to blind the researcher, and the fact that randomization enables statistical analyses. The advantages of observational studies are that they are ethically safe, less expensive, and require less administration than an RCT. RCTs often includes too few observations to investigate rare events.
The larger amount of data is a strength of register studies together with the opportunity to match study groups.
Register studies also investigate the performance or effect of an intervention in everyday practice, and not only in a specific clinical or laboratory environment. They therefore prevent performance biases and the results can often be generalized. The disadvantages of register studies are that there are no controls and that there may be hidden confounders. In observational studies such as register studies, there is also always the risk of bias:
selection bias, detection bias, reporting bias and so on (115). There can also be problems achieving completeness and response rates that are needed for sufficient analyses.
It has previously been stated that randomization is not possible in register studies, but there are now studies in which randomization occurs at registration and cluster randomization studies that also use registers. A well- performed example of a register RCT is the TASTE-study by the Swedish Coronary Angiography and Angioplasty Registry (116).
PROMs
PROs
A patient-reported outcome (PRO) is a patient’s direct self-reported health status at any time, without external interpretation. However, outcomes do not necessarily have to be reported directly after an intervention: PROs can be presented at any time and represent the individual’s valued feelings, functional ability, pain, and so on with respect to their health status at that particular moment.
PROMs
A patient-reported outcome measure (PROM) is a standardized instrument to measure PROs such as HRQoL, pain, functional impairment, or activity level. A PROM can be either generic or specific. A generic PROM
measures PROs not specific to treatment or disease and can be used to compare results across different populations and study groups, such as between different registers. The EQ-5D is an example of a generic PROM (7, 8). Specific PROMs measure constructs of health specific to a defined treatments or interventions, a disease or conditions, or specific body regions. These are important when investigating a specific outcome for a particular disease or following a treatment, but they cannot be used across areas other than those investigated. The Oxford Hip Score (117) is an example of a specific PROM.
Important factors of PROMs design
To analyze outcomes, PROMs must be valid and reliable.
Other requirements for the instrument are a design providing for the highest possible response rate but at the same time sensitive enough to detect even small changes following a treatment or intervention. There is always the risk of floor or ceiling effects if patients choose to answer in the extreme when using scales such as a VAS or EQ-5D. It is therefore important that the instrument has enough levels to prevent these effects. Implementing PROMs to investigate any changes following a treatment or intervention is a delicate task. If the applied instrument is unable to detect changes, no reliable analyses are possible. However, if there are too many questions, the response rate will also be too low for a reliable analysis.
Interpreting PROMs
There are also several considerations when interpreting PROMs, besides the risk of floor or ceiling effects. Scores from the same instrument may vary between populations.
For instance, there are several national value sets for the EQ-5D index, which is a weighted measure. These value sets are specific to the nation’s cultural norms and are based on studies of the general population using time trade-off or VAS studies. Populations may value measured areas differently because of cultural differences.
To account for these differences, national value sets weight the patient’s responses differently. As a result, comparisons of PROMs such as the EQ-5D between nations are difficult, and imply that trends, rather than exact values, need to be taken into account. It may also be difficult to decide whether changes measured following a treatment or an intervention actually are big enough to represent a clinically relevant difference.
Minimal important difference
As PROMs have become a tool that is frequently used by both clinicians and decision makers in the assessment and comparison of treatments and interventions, there is a need to determine the level of change that constitutes
a minimal important difference (MID). Several methods with similar denominations (e.g. minimal clinical important difference, MCID and minimal detectable change, MDC) confirm clinical relevance and its usefulness for implementation in clinical practice (118).
However, since these MID tools are specific to different PROMs, conditions, interventions and populations, there are no standard MIDs. MIDs therefore needs to be interpreted with caution and should take account of measurement error for the PROM. One example of this is the MDC90, which is an MDC estimate with a confidence level of 90% (119, 120). Further, MIDs calculated from individual responses may not be able to be translated into changes measured at population level.
One example is, if an MID value is established at patient level and the average change for a population is below that MID value, the distribution of change is then more important than the average change. A distribution with a narrow change probably indicates ineffective treatment.
However, a broad distribution of change indicates that treatment was probably either beneficial or harmful to some portion of the population (121). MID values can be difficult to interpret and misunderstandings of their implications for a population level compared with a patient level occur frequently. Small changes at population level are easily dismissed as clinically irrelevant when they may actually show significant differences following the treatment of a group within the larger population (122).
PREMs
Patient-reported experience measurements (PREMs) are not to be confused with PROMs. PREMs represent the patient experience of their care, not the outcome following an intervention. PREMs can provide a patient perspective on care and thus be useful for improvements at a single clinic (123), but their usefulness for national quality registers can be questioned. PREMs have not been used in any of the studies in this thesis.
The EQ-5D
One of the most commonly used health-related quality of life (HRQoL) instruments is the original three- level form of the EQ-5D, the EQ-5D-3L (7, 8). This is a short survey and it is recognized as valid in many populations and conditions, including THR populations (8, 112). The EQ-5D descriptive system includes five dimensions of health: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. Each dimen- sion is covered by one question with three levels of severity: no problems, moderate and severe problems.
The descriptive system yields 243 possible health states.
By applying weights from a specific value set, each health state can be transformed into a single index, which serves as an overall measure of HRQoL. There are different index value sets for different countries to reflect response norms for a given population.
Strengths and weaknesses of the EQ-5D
The EQ-5D questionnaire is short, making it easy to complete, thus contributing to high response rates.
The nature of the EQ-5D makes it useful not only for comparing HRQoL between populations but also for calculating cost effectiveness between treatments or interventions. There are, however, limitations to the EQ-5D. The national value sets are specific to a nation’s cultural norms based on studies of the general population using time trade-off or VAS studies.
Populations may value measured areas differently because of cultural differences. To account for these differences, national value sets weight patient responses differently. As a result, comparisons of PROMs such as the EQ-5D between nations are difficult and this implies that trends rather than exact values should be taken into consideration. Moreover, when measuring a change over time or after an intervention, floor or ceiling effects may appear. If an individual has a high EQ-5D index prior to an intervention, there is little room for improvement, resulting in a ceiling effect.
However, an individual with a low EQ-5D would have a much higher capacity for improvement. This creates strong dependence on the patient’s status prior to an intervention to measure change the intervention brings about in the patient’s HRQoL. Clustering has also been observed in the indices. For example, in the Swedish OA-population eligible for THR using the British value set, indices of 0.088 and 0.69 are very common (6).
There was no Swedish value set until 2014 when Burström et al. developed a Swedish value set using the time-trade off method (TTO) (124). For this reason, the register used the British value set. The index ranges
from a minimum value of -0.594 to a maximum value of 1.0. Negative values through 0 represent the worst possible health state and 1 represents the best possible health state. Following a study in 2015 by Nemes et al. in which they stated that the Swedish value set was better suited to a Swedish THR population than the Brittish value set (125), the SHAR changed to the Swedish value set in 2016.
The survey also comprises a vertical visual analog scale (EQ VAS), where the patient describes their total health from 0 (worst) to 100 (best). The briefness and simplicity of the survey makes it a popular instrument for assessment (7). The EQ-5D-3L is a part of the standard follow-up procedures for patients both pre- and postoperative in several arthroplasty registries (6, 8, 9, 11, 126, 127). In 2002, the SHAR started to register PROMs using the EQ-5D-3L survey (16, 107).
EQ-5D 5L
The original version of the EQ-5D, the EQ-5D-3L, has been questioned, due to profound ceiling effects, low sensitivity and the lack of descriptive richness. As it has a limited ability to measure small yet clinically relevant changes in the outcome following interventions, its usefulness in assessing interventions has been debated (128–130). These limitations have been reported for both the general population and specific patient groups (128–
132), including THR-populations (9, 10). Among THR patients, the EQ-5D-3L exhibits particularly difficulty in assessing outcome in the mobility dimension, where the options “no problems”, “some problems” and “confined to bed” limits its use in describing the limitations in mobility commonly experienced by patients with hip disorders. These patients typically experience limping, a limited range of hip joint motion, impaired walking capacity and often require different aids for mobility, but they are seldom confined to bed. Similarly, the response levels of self-care and usual activities (“no problems”,
“some problems” and “unable”) limit the range of responses for individuals with moderate to severe disability (10, 133, 134).
So, the EuroQol group has developed a new version of the questionnaire, the EQ-5D-5L, offering respondents five levels of responses instead of three: no, some, moderate, severe and extreme problems (11). In the three-level survey, a response of “no problems” in all dimensions would be notated as 11111 and “severe problems” as 33333. For the five-level survey, “no problems” in all dimensions would also be notated as 11111, but a response of “extreme problems” in all
five dimensions would be notated as 55555 to reduce the floor and ceiling effects. This gives the new five- level version 3,125 unique health states instead of 243 in the three-level version. The idea is that the increased number of response levels will provide a better profile of the patient’s health. The EQ-5D-5L instrument has been compared with the 3L version in several studies and been reported to be valid, to reduce ceiling effects and to increase discriminatory power in several populations (132, 135–137), as well in THR populations (9, 10). For those populations (for example Sweden’s) where no five-level value sets are available to calculate an index score, “crosswalk”-algorithms are available from 3L to 5L (11). The 5L version is yet to be tested and validated in a Swedish THR-population.
EQ-VAS
The second part of the EQ-5D contains a VAS addressing general health (EQ VAS) (112), where 0 and 100 represent the worst and best possible health state, respectively. The floor or ceiling effects and the multimodal distribution of indices have made the EQ- 5D subject to criticism (138–144). These facts need to be taken into consideration when performing statistical analyses of the EQ-5D. Despite this, the EQ-5D is a useful tool for measuring patient HRQoL pre- and post THR surgery.
Pain VAS
The pain VAS (110) ranges from 0 to 100, where 0 represents no pain and 100 the worst possible pain.
In the SHAR, the hip pain VAS is registered pre- and postoperatively at one, six and ten years (16, 107). In the Swespine the leg and back pain VAS is registered pre- and postoperatively at one, two, five, and ten years (2).
Oswestry Disability Index
The Oswestry Disabilty Index (ODI) is frequently used to measure the degree of disability and estimate quality of life in a patient with low back pain. The self- completed questionnaire contains ten topics concerning intensity of pain, lifting, ability to care for oneself, ability to walk, ability to sit, sexual function, ability to stand,
social life, sleep quality, and the ability to travel. Each topic’s category is followed by six statements describing potential scenarios in the patient’s life relating to the topic. Each question is scored on a 0–5 scale with zero indicating the least disability and five indicating the most severe. The scores for all questions are then summarized and multiplied by two to obtain the index, ranging from 0 to 100. Zero is equated with no disability and 100 is the maximum possible disability (113). In the Swespine the ODIs are registered pre- and postoperatively at one, two, five, and ten years (2).
Other measurements used in these studies
Satisfaction
In the SHAR satisfaction with the outcome following THR was measured by the satisfaction VAS. This scale ranges from 0 to 100, where 0 represents very satisfied and 100 very dissatisfied. In the SHAR, satisfaction VAS is registered postoperatively at one, six and ten years (16, 107). In the Swespine outcome satisfaction following LSS is measured by the patient rating the experienced outcome using a three-level questionnaire: 1. Satisfied, 2. Uncertain, 3. Dissatisfied. Satisfaction is registered at one, two, six and ten years following surgery (2).
Charnley class
The Charnley classification is a patient-reported survey in the SHAR, but was originally developed to be assessed by an observer, such as the orthopedic surgeon.
The classification determines the musculoskeletal comorbidity status of the patient according to: Class A, corresponding to a unilateral hip disorder, Class B, a bilateral hip disorder, and Class C, a walking impairment due to multiple joint involvement or other medical comorbidities (111). The question of whether to divide Class B into two groups, those with one side or the other already treated, has been discussed (145). However, the evidence for this is not yet strong enough for a new fourth class. In the SHAR, the Charnley classification is registered preoperatively (16, 107).
The overall objective of the studies presented in this thesis was to investigate the clinical utility of PROMs following THR and LSS. Specifically, the aims were to:
• Explore patient-reported outcomes following THR and LSS in patients in whom both procedures have been performed and to compare these results to patients in whom only one of the procedures was performed. (Paper I and II)
• Explore whether the order of THR and LSS procedures in patients in whom both procedures have been performed within a short period of time of two years influences the patient-reported outcome following the last procedure. (Paper III)
• Investigate the opportunity to use PROMs one year following THR in order to predict the risk of a late reoperation following THR. (Paper IV)
• Calculate for the 3L and 5L-verions of the EQ-5D an estimate for the different response options by dimension using the EQ VAS. To assess the measurement properties of the EQ-5D-5L and investigate any differences compared with the EQ-5D-3L, preoperatively and one year postoperatively in a Swedish THR-population. (Paper V)
Aims
Patients and methods
Patients
The data used for papers I, II and III were retrieved from the SHAR and the Swespine for patients undergoing surgery from 2002–2012. These years were selected since the SHAR first started to collect PROMs in 2002. After linking of the registers using the PIN as a common identifier, selections were made based on predefined selection criteria specific to the respective studies. For each study, a separate dataset was compiled.
In paper IV, demographic, surgical and PROMs data were retrieved from the SHAR covering patients undergoing primary THR in 2002–2015. In order to reduce confounding, patients included in the analysis were selected according to preset criteria. In paper V, all patients were recruited from one of the seven publicly funded hospitals in western Sweden in 2015. A separate database was established for the EQ-5D-5L versions of the surveys while the EQ-5D-3L versions of the survey
were retrieved from the regular PROMs database. All the data sets that were used have been stored within the highly protected IT infrastructure of the SHAR at the Register Centre of Västra Götaland Region.
Table 1 summarizes the number of patients included in the different studies. A clarification is needed to explain the differences regarding patients and procedures between Papers I and II-III. Due to a misunderstanding between authors and statisticians in Paper I the SHAR dataset was expanded until 2013 and the Swespine dataset from 1998 to 2014. However, only patients with LSS prior to THR were included. As a result, only 86 patients with surgery outside the selected years, 2002–2013 were included in the study population in Paper I. This mistake was discovered after publication. These extra patients did not affect the outcome or analysis of the studies. This has been carefully investigated and results without these 86 patients are presented in Table 5.
Table 1
Tot. no of procedures Tot. no of patients Year of surgery Comments Paper I
SHAR Swespine Study group Matched group
139,697
47,433 109,306
43,767 997 997
2002–2013 1998–2014
Study and control group following selection Fig 3
Paper II SHAR Swespine Study group Control group
159,247
34,559 126,752
25,394 220 220
2002–2012 2002–2012
Study and control group following selection Fig 5
Paper III SHAR Swespine Study group
159,247
34,559 126,752
25,394 255
2002–2012 2002–2012
Study group following selection Fig 6
Paper IV SHAR Study population -number reoperated
141,300 75,899 1,405
2002–2014
All THRs in Sweden 2002–2014, study population selection Fig 7
Paper V THRs Preop 3L Preop 5L Postop 3L Postop 5L
1,567 1,182 767 1400 508
2015 2015 2015 2016 2016
All patients eligible for THR in western Sweden in 2015 were intentionally invited to take part in the study
Ethical considerations
According to the Patient Data Act (SFS2008:355), the collection of data in Swedish national quality registers does not require written or oral consent from patients.
The law obliges the health-care provider to inform patients that data will be registered and used for quality improvement and research and that they may opt out at any time and have their data deleted from the register.
Written information about the collection of data for quality registers is normally provided before or at the preoperative visit. Ethical review board approval is required for all research using national quality register data. Information on ongoing research projects is provided by the respective registers and is ideally posted on the register webpages.
For Paper V, patients were recruited from all public hospitals performing THRs in the Västra Götaland Region in 2015. Information about the study was provided before or at the pre-operative visit by each participating unit. Repeat study information was included when follow-up questionnaires were sent out.
The return of questionnaires was regarded as consent to participate. Information about the study was also available at the SHAR website.
For all studies presented in this thesis, ethical review board approval was obtained from the Regional Ethical Review Board in Gothenburg, Sweden (Papers I-III entry number 236-13, Paper IV entry number 368-17, and Paper V, entry number 516-14).
Methods
Papers I and II
Papers I and II generally had a similar study structure.
Following linkage of the SHAR and the Swespine a set of predefined step-wise selection criteria was applied in both studies (Figures 3 and 5). Following these steps a study population containing a study group and control group was created. In Paper I the influence of a previous LSS on PROMs one year following THR compared with a group with only THR was investigated. In Paper II the opposite, PROMs one year following LSSS in patients with and without a THR prior to LSSS, was investigated.
In Paper I there was an additional investigation were the prevalence of a prior LSS among patients with THR due to hip OA during 2012 was calculated (Figure 4).
Both studies are prospective observational register studies presenting the outcome without drawing conclusions on causality.
Paper III
Paper III is similar to Papers I and II in terms of study structure. As in the previous papers, the SHAR and Swespine were linked and a set of predefined selection criteria was then applied in a step-wise fashion (Figure 6). The study group consisted of patients who had undergone both THR and LSSS within a two-year period. We investigated whether the order of surgery influenced PROMs (EQ-5D and EQ VAS) one year after the last procedure.
As for Papers I and II, this study is a prospective observational register study presenting the outcome without drawing any conclusions on causality.
Paper IV
For Paper IV data from SHAR were obtained and a set of predefined step-wise selection criteria was applied (Figure 7). From the selected study group a study population of those patients with a late reoperation was identified. The outcome was reoperation for all reasons and all types of surgical procedures later than one year after the index surgical procedure. An investigation was then made of the opportunity to predict the risk of late reoperation. The predictors of reoperation were age, sex and PROMs collected from the PROMs program in the SHAR (146), which includes the EQ-5D health status questionnaire (109), a hip pain visual analogue scale (VAS) (110) and, at follow-ups, satisfaction with the outcome using a VAS.
A patient-reported Charnley classification used to determine patient-reported musculoskeletal comor- bidity was also included (111).
Paper V
In order to study the aim and research questions in Paper V, the aim was to invite all patients eligible for THR in 2015 on the basis of primary hip osteoarthrosis at any of the seven publicly funded hospitals performing THRs in the western region of Sweden (Västra Götalandsregionen). During the standard preparatory preoperative visit prior to the THR procedure patients regularly completes the EQ-5D-3L questionnaire as part of the routine PROMs program of the SHAR (147).
Two weeks prior to this visit, invitation letters including preoperative information are sent to the patients by each hospital’s waiting list coordinators. During the study period, these letters also included the 5L version
